Williams on behaviorism

CSGnet Archive CSG2004
1

[From Bruce Abbott (2004.08.19.1105 EST)]

>Bill Williams, various recent posts (see below):

Bill, recently you've offered a series of assertions concerning a topic I
have a great deal of familiarity with (the literature of "learning and
behavior"). These assertions of yours are copied into the list below:

1. As far as I know the psychologists never actually got around to
defining either a "stimulus" or a "response." (15 August 2004 11:10 PM CST)
2. Did John Watson knowingly tell a fib about his experiments with "Little
Albert?" If that is there really was a "Little Albert." (16 August
2004 1:40 AM CST)
3. I thought it was well known that the advocates of behaviorism never
defined either a "stimulus" or a "response." (17 August 2004 9:40 AM CST)
4. And, I am sure that Rick has encountered the argument that the John
Watson's whole "little Albert" experiment was a sheerly fictional creation.
If we are going to extend the concept of modeling to fiction, may I vote to
include Charles Dickens? David Copperfield is I think probably more real
that John Watson's "little Albert."
(17 August 2004 9:40 AM CST)
5. Since, as far as I know behaviorism never defined what a stimulus was,
there never was a stimulus to observe. (18 August 2004 7:30 PM CST)
6. Since a stimulus or response are both undefined, there was to begin
with no observed relationship which the organism was "mediating." (18
August 2004 8:00 PM CST)
7. Those who are familiar with the critical literature of experimental
psychology may recall the often cited, over a period of decades, experiment
of a cat escaping from a puzzle box. . . . Eventually when the experiment
was repeated, a really good thing to do, it was noted that a cat in the
puzzle box only escaped when the cat could see a person. What was happening
was not that the cat was attempting to escape from the box, but rather that
the cat was behaving in a way that was a disiplaced greeting. The people
who had originally "seen" and reported the cat's behavior as a response to
an escape puzzle had been mistaken. (18 August 2004 9:00 PM CST)

The questions I have for you is, from where are you getting the information
on which you base these assertions?

Regards,

Bruce A.

2

From[Bill Williams 20 August 2004 4:50 AM CST]

[From Bruce Abbott (2004.08.19.1105 EST)]

>>Bill Williams, various recent posts (see below):

Bill, recently you've offered a series of assertions concerning a topic I
have a great deal of familiarity with (the literature of "learning and
behavior"). These assertions of yours are copied into the list below:

Let's start with # 7 in your listing, in which I said:

7. Those who are familiar with the critical literature of experimental
psychology may recall the often cited, over a period of decades, experiment
of a cat escaping from a puzzle box. . . . Eventually when the experiment
was repeated, a really good thing to do, it was noted that a cat in the
puzzle box only escaped when the cat could see a person. What was happening
was not that the cat was attempting to escape from the box, but rather that
the cat was behaving in a way that was a disiplaced greeting. The people
who had originally "seen" and reported the cat's behavior as a response to
an escape puzzle had been mistaken. (18 August 2004 9:00 PM CST)

My reference for # 7 is the following copied from:

De Wall, Frans B. M 1999 "The Pitfalls of not knowing the whole Animal"

  The Chronicle of HIgher Education 5 March p. b4

I came across an amusing illustration in the scientific literature

of the pitfalls for those who fail to pay attention to the whole

animal. Bruce Moore and Susan Stuttard psychologists at Dalhousie

University in Nova Scotia, reported in Science 1979 that they had

tried to replicate a 1946 study widely cited as demonstrating the

ability of cats to work their way out of a puzzle box, a container

whose door was operated by moving a rod. The earlier study done

by Edwin R. Gurthrie and George P. Horton documented in great detail

how cats rubbed against the interior of the box with steriotyped

movements. In the process they moved the rod and escaped. Guthrie

and Horton had deemed it highly significant that all the cats in

the experiement showed the same rubbing pattern, which they

believed they had taught the animals through the use of food as

rewards: This proved the power of conditioning.

    When Moore and Stuttard repeated the experiment, their cats

behavior struck them as nothing special. The cats performed the usual

head-rubbing movements that all felines--from house cats to Jaguars--

use in greeting and courting each other. Domestic cats often redirect

these movements to inanimate objects, such as the legs of a kitchen

table. Moore and Stuttard showed that the food rewards were absolutely

irrelevant: The only meaningful factor for the cats in the box was

the visiblity of people. Without training, every cat who saw

people while in the box rubbed its head, flank and tail against

the rod and got out of the box. Cats who didn't see people just

sat there. Instead of a learning experiment, the 1946 study had

been a greeting experiment. p. B5.

The questions I have for you is, from where are you getting the

information on which you base these assertions?

Maybe the thing to do is for me to do would be obtain the citation for

the Moore and Stuttard paper in Science. I'm not sure now, but I seem

to remember talking to Greg Williams about the Moor and Stuttard paper

so perhaps I learned of the paper from Greg.

My understanding, and I would be interested in your assessment, is that

in the original puzzle box experiement food was thought to be the

stimulus. The nominclature seems a bit confused in my view because the

term "reinforcement" is also used. In any case whether as a "stimulus"

or as a "reinforcement" Gurthrie and Horton identified food as the

external causal influence that prompted the cats to escape from the

puzzle box. And, apparently the study was widely cited for a third of

a century-- until Moore and Stuttard in their replication found that

what the cats were doing was instead a displaced greeting behavior.

As I understand it, in conducting their experiment Gurtherie and

Horton "saw" food as a "causal force" and reported this relationship

in terms of a "factual observation." Actually, however, the causal

relationship is, as Moore and Stuttard, found what the cat is doing

has to be understood in a scheme that is quite different than the

interpretation provided using the theretical scheme provided by

behaviorism.

When in recent posts Rick Marken claimed that he could "see" a stimulus,

I thought of the Gurthrie and Horton experiment and the reinterpretation

of what the cats were doing by Moore and Stuttard. It appeared to me that

the argument that Rick was developing-- that he could "see" a stimulus was

not consistent with his earlier argument that ?" You can't tell what

someone is doing, by watching what they are doing."? Nor, was it

consistent with what I had thought was the critical/skeptical background

to the notion of a "test" for a controlled variable.

It is possible that De Wall's description is a bit garbled, however, in

my view it supports a rejection of claims that it is obvious in an

experimental situation what is the stimulus or the reward. What people

may have thought to have been the stimulus or the reward have, when the

experiments have been replicated turned out to be as De Wall says to be

"irrelevant."

I made a somewhat similar mistake. I owned a Belgan Shepherd "Snips."

Snips was a lot more agressive than a typical German Shepherd. Our next

door neighbor's girl of about 5 had an enourmous tomcat named Kitty. One

day I heard Snips barking, and the cat howling. Carol was standing in

our back yard with Kitty in her arms and Snips was trying to get at

Kitty. After calling off Snips, and getting Carol to let loose of Kitty,

I told Carol that I didn't think Snips and Kitty were ever going to be

pals. And, it wasn't a good idea to try to get Kitty and Snips together.

Somebody might get hurt. Carol was initially a bit sullen. Her parents

could be rather strict and she was anticipating that I was going to scold

her, or tell her folks what she'd been upto. But, she was also a bit

independent, and when I attempted to explain about Snips and Kitty, she

looked at me like-- all men are just dumber than dirt and she said, "I

know that." I was a bit puzzled, and I asked her, "So why were you

trying to get Kitty and Snips together?" Her answer was, "Oh, I just

like to see Kitty's tail get big."

Now, maybe Rick can "see" a stimulus better than I can, but I doubt that

even an expert hero of PCT sophistology would have been able to "see"

that what was going on in my back yard when Carrol set about creating a

snarling, howling furball was the result of Carrol's being reinforced by

seeing Kitty's tail get big.

Bill Williams

3

[From Bruce Abbott (2004.08.20.2050 EST)]

Bill, for some reason your typing in your reply came out double
spaced. I’ve removed the extra spaces in the quoted areas below.
Perhaps there is a setting on your email program that can be changed to
fixed this.

[Bill Williams 20
August 2004 4:50 AM CST]

[From Bruce Abbott (2004.08.19.1105 EST)]

Bill Williams, various recent posts (see below):

Bill, recently you’ve offered a series of assertions concerning a
topic I

have a great deal of familiarity with (the literature of
"learning and

behavior"). These assertions of yours are copied into the
list below:

Let’s start with # 7 in your listing, in which I said:

  1. Those who are familiar with the critical literature of
    experimental

psychology may recall the often cited, over a period of decades,
experiment

of a cat escaping from a puzzle box. . . . Eventually when the
experiment

was repeated, a really good thing to do, it was noted that a cat in
the

puzzle box only escaped when the cat could see a person. What was
happening

was not that the cat was attempting to escape from the box, but
rather that

the cat was behaving in a way that was a disiplaced greeting. The
people

who had originally “seen” and reported the cat’s behavior
as a response to

an escape puzzle had been mistaken. (18 August 2004 9:00 PM
CST)

My reference for # 7 is the following copied from: De Wall, Frans B. M
1999 “The Pitfalls of not knowing the whole Animal” The
Chronicle of HIgher Education 5 March p. b4
. .
.
My understanding, and I would be
interested in your assessment, is that

in the original puzzle box experiement food was thought to be the

stimulus. The nominclature seems a bit confused in my view because
the

term “reinforcement” is also used. In any case whether as a
“stimulus”

or as a “reinforcement” Gurthrie and Horton identified
food as the

external causal influence that prompted the cats to escape from the

puzzle box. And, apparently the study was widely cited for a third
of

a century-- until Moore and Stuttard in their replication found that

what the cats were doing was instead a displaced greeting
behavior.

As I understand it, in conducting their experiment Gurtherie and

Horton “saw” food as a “causal force” and reported
this relationship

in terms of a “factual observation.” Actually, however,
the causal

relationship is, as Moore and Stuttard, found what the cat is doing

has to be understood in a scheme that is quite different than the

interpretation provided using the theretical scheme provided by

behaviorism.

Interesting. I retrieved the de Waal (1999) article you quoted from
as well as the Moore and Stuttard (1979) article in Science that
was the basis of DeWall’s remarks. Frans de Waal is an ethologist
(zoologist specializing in the study of animal behavior); Bruce Moore was
and is a comparative psychologist with much the same focus. (I assume
that Susan Stuttard was working with Moore as a graduate student at the
time their article was written.) It would appear from their
statements in these articles that none of these folks had more than a
passing acquaintance with the subfield within psychology called
“learning and behavior.”
In their Science article, Moore and Stuttard press the thesis that
“behaviorists” have conducted their research “without
regard for the natural behavior of the animals used as
subjects.” To support this thesis, they first assert that
“The pigeon’s “operant” key-pecking response . . . was not
identified as a simple grain-pecking reaction, and a generation of
investigators therefore believed that they had taught tens of thousands
of pigeons, individually, how to peck.” This is sheer
nonsense, on three counts. First, pigeon response-keys were
developed to take advantage of the pigeons natural pecking behavior.
Secondly, although pigeons naturally peck at seed-like objects on the
ground, they do not naturally peck at response-keys, which are mounted
high on the chamber wall (at about beak-level when the pigeon is standing
erect). Instead, the key-peck must be shaped into existence through
the use of differential reinforcement of successive approximations to the
target behavior. Moore and Stuttard apparently were not aware of this,
but relied instead on reports of the results of a procedure developed by
Brown and Jenkins called “autoshaping,” which produces
keypecking automatically through a process that was quickly recognized as
capitalizing on Pavlovian conditioning of the pigeon’s innate pecking
behavior. Thirdly, investigators using the standard shaping procedure
never believed that they were teaching their pigeons how to peck.
What their procedure was aimed at accomplishing was teaching the pigeon
what to peck in order to receive grain.
Moore and Stuttard provided additional support for their thesis in the
form of their replication of the Guthrie and Horton (1946) study, which
itself followed up on Thorndike’s famous “cats in the
puzzlebox” experiments. The Guthrie and Horton study used a box
equipped with a pole mounted vertically in the center and articulated in
such a way that tipping the pole in any direction would open the door to
the box. A camera recorded the cat’s position at the moment the pole was
tipped. Fifty-two cats were tested, and each eventually developed a
stereotyped way of pushing against the pole that was repeated with very
little change from trial to trial. Moore and Stuttard claimed that
Guthrie and Horton were unaware that the pole-related behaviors observed
in their cats were species-typical behaviors. In their own
replication, the cats performed these actions even in the absence of
“reinforcement” and, when people were not visible, many of them
simply laid down in the box. They concluded that the cats in the Guthrie
and Horton experiment had learned nothing; they were simply expressing
instinctive behaviors elicited by the presence of the observers.
The problem here is that Moore and Stuttard missed the whole point of the
Guthrie and Horton experiment, which was to observe precisely what the
animal was doing at the moment its actions triggered release from the
box, and how that action changed over trials. Although the cats
initially may have been rubbing against the pole for the reason Moore and
Stuttard assert, each cat ended up, after many trials, performing the
behavior in a highly stereotyped way, in contrast to the variability seen
initially. Moore and Stuttard did not provide an extrinsic reward
for pushing against the pole, and their cats continued to push against
the pole in highly variable ways across trials.
Thorndike’s (1898) puzzlebox experiments employed a variety of boxes
equipped with a variety of latch-mechanisms that in general did not
provide the occasion for “rubbing” behavior. Thorndike,
however, noted that many of the behaviors observed in the puzzlebox may
have been instinctive, or components of instinctive behaviors, although
this is not mentioned by Moore and Stuttard. And in a footnote to their
article, even they admit that “Thorndike’s original monograph also
dealt with the reactions of cats in puzzle boxes, but his apparatus and
techniques were very different from those of Guthrie and Horton, and his
data were not compromised.” In another footnote, they added
the following caveat: "We have not, of course, set out to
demonstrate that learning was everywhere irrelevant to the behavior of
Guthrie and Horton’s cats. We have argued only that their principal data
do not constitute evidence of learning by contiguity,
“superstition,” or any other process. Yet in his
Chronicle article, DeWaal concludes that “Instead of a
learning experiment, the 1946 study had been a greeting
experiment.”

De Waal is correct that “behaviorists” have often ignored the
typical behaviors of their animals as observed under various
circumstances, he does not seem to be aware that “behaviorists”
have been aware of the need to take these into account for several
decades now, beginning with Frank Beach’s famous “The Snark was a
Bojum” article in the 1940s and more recently including Breland and
Breland’s (1960) “The Misbehavior of Organisms,” Robert Bolles’
“Species-Specific Defense Reactions,” and the Brown and Jenkins
autoshaping paper. He cites B. F. Skinner’s “Case History in
Scientific Method” paper for asserting that “Pigeon, rat,
monkey, which is which? It doesn’t matter” (referring specifically
to the demonstrable fact that all produce the same distinctive patterns
of behavior characteristic of the various schedules of reinforcement) as
if Skinner did not recognize that each species has its own distinctive
capabilities, apparently unaware that Skinner (and Watson before him)
recognized the important selective influence of the evolutionary
process.

Now, what about those others I asked about? Where did you encounter
those?

Regards,

Bruce